Research On Collective Spatial Keyword Query And Recommendation

With the popularization of mobile smart devices and the rapid development of mo-bile Internet,location-based services have gained popularity.In recent years,searching and recommendation techniques have become hot research area in the field of spatial database.Spatio-textual object search can make comprehensive use of spatial proximity and textu-al relevance for query processing to meet the diverse query needs of users.Spatio-textual object recommendation recommends new places to users by analyzing their personalized preferences.Spatio-textual object searching and recommendation have important research value in helping merchants to allocate resources,excavate potential customers and help cus-tomers to make decisions.Based on the common requirements of location-based services,we study efficient query and personalized recommendation technology for spatio-textual objects to support more ef-ficient and accurate spatial-temporal retrieval.In terms of spatio-textual object retrieval,we firstly design a collective spatial keyword query framework with semantic fusion to solve the problem that the existing methods are not capable of supporting multi-objective query.It uses the method of word embedding to analyze the semantics of the text.Then we use a hier-archical index structure LIR-tree to organize data with seamless fusion of multi-dimensional information.Based on this,a series of bounded theorems with theoretical guarantees are used to optimize the query processing to achieve effective query.It ultimately ensures the efficient processing of single or multi-objective retrieval at the semantic level.In terms of personalized recommendation of spatio-textual objects,we implement a recommendation algorithm integrating spatial and temporal characteristics to effectively capture user prefer-ences and solve the data sparsity problem of user behaviors.The spatio-textual objects searching and recommendation have important theoretical and practical value.The correlation algorithms are compared and verified by experiments on real datasets,and the experimental results prove the feasibility and effectiveness of the correlation searching and recommendation algorithms.In addition,we combine the col-lective spatial keyword query with POI recommendation to build a query-recommendation prototype system.